Package 'dsPUcopulaClient'

Title: Client Utilities for Partition-of-Unity Copula Modelling in DataSHIELD
Description: Provides client-side helper functions for interacting with the dsPUcopula DataSHIELD server package. The helpers wrap the remote procedures for fitting partition-of-unity copula models, estimating marginal distributions, simulating copula-based samples and retrieving synthetic data in a privacy-preserving manner.
Authors: Andreas Mändle [aut, cre]
Maintainer: Andreas Mändle <[email protected]>
License: MIT + file LICENSE
Version: 0.1.0
Built: 2026-05-20 10:35:04 UTC
Source: https://github.com/bips-hb/dsPUcopulaClient

Help Index

Estimate marginal distributions inside DataSHIELD

Description

Computes marginal distribution estimates for the variables identified by the server-side symbol x through the estimateMarginalsDS helper. The result is stored remotely as marginals_list for later reuse.

Usage

ds.estimateMarginals(x, datasources = NULL, method = "ecdf")

Arguments

x

character(1). The name of the server-side object that holds the variables for which the marginals should be estimated.
datasources

Optional list of DataSHIELD connections as returned by DSI::datashield.connections_find().
method

character(1). The marginal estimation approach to use. The default is "ecdf", but the value must match a method supported on the server side.

Value

Invisibly returns TRUE once the remote object has been created.

Examples

## Not run: 
ds.estimateMarginals("PU_copula_model$data")

## End(Not run)

Fit a Partition-of-Unity Copula Model on the server

Description

Creates or replaces the PU_copula_model object in the connected DataSHIELD sessions by delegating to the server-side fitPUcopulaDS function.

Usage

ds.fitPUcopula(
  data,
  driver_strength_factor = 0.5,
  bin_size = 3,
  datasources = NULL,
  ...
)

Arguments

data

character(1). The name of the server-side data frame that contains the copula samples.
driver_strength_factor

numeric(1). Tuning parameter for the driver strength estimation. Values must be positive.
bin_size

integer(1). Bin size used during the density estimation steps of the algorithm.
datasources

Optional list of DataSHIELD connections. When omitted the currently active connections obtained via DSI::datashield.connections_find() are used.
...

Additional arguments passed through to the server-side fitPUcopulaDS function.

Value

Invisibly returns TRUE once the remote object has been created.

Examples

## Not run: 
ds.fitPUcopula(data = "D", driver_strength_factor = 0.6, bin_size = 4)

## End(Not run)

Retrieve synthetic data from the server

Description

Aggregates the synthetic observations generated by the server-side generateSyntheticDS function and returns them to the client.

Usage

ds.generateSynthetic(n, datasources = NULL)

Arguments

n

integer(1). Number of synthetic observations to retrieve.
datasources

Optional list of DataSHIELD connections to use.

Value

A data frame with n rows containing the synthesised observations.

Examples

## Not run: 
synth <- ds.generateSynthetic(500)

## End(Not run)

Simulate from the fitted copula model

Description

Generates a simulated sample of size n from the copula model previously stored in PU_copula_model. The simulated uniform scores are kept on the server-side as PU_copula_model_u_sims_clientcode.

Usage

ds.simulateCopula(n, datasources = NULL)

Arguments

n

integer(1). Number of draws to request from the server.
datasources

Optional list of DataSHIELD connections to use.

Value

Invisibly returns TRUE once the simulated sample has been stored on the server.

Examples

## Not run: 
ds.simulateCopula(1000)

## End(Not run)